Available online at www.sciencedirect.

com

Intelligence 36 (2008) 127 – 142

Relevance of education and intelligence at the national level

for the economic welfare of people
Heiner Rindermann ⁎
Institute of Psychology, Otto-von-Guericke-Universitaet Magdeburg, PO Box 4120, D-39016 Magdeburg, Germany
Received 7 March 2006; received in revised form 12 February 2007; accepted 12 February 2007
Available online 27 April 2007

Abstract

Cognitive abilities are important for the economic and non-economic success of individuals and societies. For international
analyses, the collection of IQ-measures from Lynn and Vanhanen was supplemented and meliorated by data from international student
assessment studies (IEA-Reading, TIMSS, PISA, PIRLS). The cognitive level of a nation is highly correlated with its educational
level (r = .78, N = 173). In international comparisons, it also shows a high correlation with gross domestic product (GDP, r = .63,
N = 185). However, in cross-sectional studies, the causal relationship between intelligence and national wealth is difficult to
determine. In longitudinal analyses with various samples of nations, education and cognitive abilities appear to be more important as
developmental factors for GDP than economic freedom. Education and intelligence are also more relevant to economic welfare than
vice versa, but at the national level the influence of economic wealth on cognitive development is still substantial.
© 2007 Elsevier Inc. All rights reserved.

Keywords: Education; Intelligence; Economic growth; International cognitive ability comparisons; Modernity

1. The relevance of cognitive abilities at the
individual and national level
Thinking ability, knowledge and the intelligent use of
knowledge are considered to be important determinants
of life success. The correlates of intelligence at the in-
dividual level include aspects of civil life such as
employment status, income, life expectancy, moral
development, behavior complying with legal standards,
married life and beneficial education of children (Armor,
⁎ Current address: Karl-Franzens-University Graz, Institute of
Psychology, Universitätsplatz 2, A-8010 Graz, Austria.
E-mail addresses: heiner.rindermann@gse-w.uni-magdeburg.de,
heiner.rindermann@uni-graz.at.
0160-2896/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.intell.2007.02.002
2003; Gottfredson & Deary, 2004; Herrnstein & Murray,
1994; Murray, 1998; Piaget, 1932; Schmidt & Hunter,
1998; Whalley & Deary, 2001). Education shows similar
positive effects (Barnett, 1998), even lowering the risk of
HIV infection (Sanderson, 2004; UNESCO, 2004).
Throughout the past century education has been
viewed as an important determinant of the cultural,
political and social success of nations and of cultural and
religious groups (macro-social level; e.g. Weber, 1988/
1920). At least since OECD studies about student
cognitive competences like PISA and the international
collection of intelligence test results by Lynn and
Vanhanen (2002, 2006) cognitive abilities are consid-
ered to be positive determinants of national wealth. The
basic assumption of international ability comparisons is
that at the macro-social level cognitive abilities are
128 H. Rindermann / Intelligence 36 (2008) 127–142
important causal factors for economic welfare (mea-
sured as gross domestic product, GDP) and perhaps also
for non-economic welfare (democracy, rule of law,
human rights, health).
Cognitive abilities help individuals to succeed in
school and find better opportunities in jobs and private
life (door opener function), and they help in everyday
life situations, especially when the understanding and
effective use of causal relationships are required
(problem solving function). Cognitive abilities are
especially important in complex occupations with high
demands on learning, novel problem solving and
independent decision-making. However, the causal
relationship is bidirectional since education and com-
plex tasks are known to raise intelligence and knowl-
edge (Ceci, 1991; Schooler, Mulatu, & Oates, 1999).
Intelligence and knowledge enhance individual and
cultural rationality. They support rational decisions
not only by private individuals, but also in institutions
and the political system. They help to assess opportu-
nities and risks (insight, precaution), to use effective
means, to achieve institutional goals, to avoid accidents,
to assess the importance of resources and anticipated
outcomes (future outlook), and to maintain a climate of
reason in which disputes are settled by verbal arguments
rather than by the use of violence or coercion (e.g.
Gottfredson, 2003; Habermas, 1976; Piaget, 1932;
UNESCO, 2004). At the level of institutions and of
nations, aggregated individual intelligence effects and
genuine macro-social intelligence effects like efficient
economic and administrative structures (“intelligent
culture”) come together.
2. Research at macro-social data level
Research about intelligence at the national level is
confronted with problems of data quality: At a more
fundamental level, the comparability of cognitive test
results across national and cultural boundaries has been
called into question (Greenfield, 1997), with national IQ-
measures being especially controversial. Lynn and
Vanhanen (2002) collected intelligence test results from
81 countries and transformed the results into one
common norm (“Greenwich IQ”). For 104 additional
countries intelligence was estimated by using data from
neighboring countries with similar social, cultural and
racial characteristics. Their collection is a landmark, but
subjected to some problems (see Barnett & Williams,
2004; Ervik, 2003; Hunt & Wittmann, in press; Richards,
2002; Volken, 2003): (1) there are countries with very
small test samples; (2) in some cases, samples seem to be
unrepresentative; (3) different tests were used across
different countries; (4) different authors conducted the
test surveys; (5) measurements of intelligence stem from
different years (Lynn and Vanhanen corrected this); (6)
missing data were estimated by using unweighted
arithmetic means of neighboring countries with similar
people. It is therefore not surprising that some researchers
are skeptical. Hunt and Sternberg (2006), for example,
criticized the “inadequate estimates of national IQ” (p.
131), and maintained that the “concept of national IQ is
meaningless without carefully designed probability
samples of the population” (p. 133). In response to
these criticisms, Templer and Arikawa (2006, p. 139)
admit problems but refer to the dearth of better
information: “There is, however, no other international
aggregation of IQs.”
But “intelligence tests” are not the only cognitive
ability tests. International student assessments (PISA,
TIMSS, PIRLS, IEA-Reading) use carefully standardized
methods of data collection, but evidence other problems:
(1) data exist mainly for developed countries; (2) they are
chiefly representative for school-aged children and
adolescents; (3) participation rates differ across countries;
(4) and school dropouts are not represented (only the part
of youth that attends school). (5) There are inconsisten-
cies in the age levels; the TIMSS and PIRLS assessments
in particular used class-level samples with different ages
across countries. However, despite the many short-
comings of IQ tests and school assessments, the Lynn
and Vanhanen IQs correlate so highly with the results of
international school assessments that the two types of
tests appear to measure the same or at least a similar
construct (Rindermann, in press).
3. Results of intelligence tests and student
assessment tests as indicators of one common
cognitive ability
There is no theoretically important difference between
intelligence tests and student assessment tests. Both
assess thinking and knowledge. Theoretically and
empirically they are similar (Coleman & Cureton,
1954; Cronbach, 1984; Ceci, 1991; Jensen, 1998). At
the national level the correlations between the two types
of test are very high (r = .80–.90; Rindermann, in press).
At the individual level, the correlation of the SAT
(Scholastic Aptitude Test or Scholastic Assessment Test)
with the ASVAB (Armed Services Vocational Aptitude
Battery) is r = .82, and with APM (Advanced Progressive
Matrices) it is r = .48 (r = .72 corrected for restricted
range; Frey & Detterman, 2004). The GCSE (General
Certificate of Secondary Education, national examina-
tion in England for students at age 16) correlates about
 H. Rindermann / Intelligence 36 (2008) 127–142
r = .69 (at latent level rl = .81) with the CAT (Cognitive
Abilities Test) and about r = .66 with the figural, school-
distant scale of the CAT (Deary, Strand, Smith, &
Fernandes, 2007).
But there is a theoretical difference between thinking
and knowledge, although these two intertwined ability
components are difficult to separate. Knowledge is
always required to solve the kinds of task that
individuals are confronted with in everyday life or that
are used in cognitive ability tests. And thinking ability
helps to increase and use knowledge.
Since intelligence tests and school assessment tests
are alternative measures of an homogeneous construct,
we can form a summary score from the two types of test
(see Methods section). This will reduce the described
flaws of the two types of test studies and will provide a
concise measure of knowledge-reduced intelligence, of
(as valuable and true estimated) knowledge and of
thinking skills that are needed to make use of this
knowledge. Being based in part on assessments of
school-aged children and adolescents, it will represent
mainly the intellectual level of young people, although it
is likely to parallel quite closely the intelligence of
adults in the country.
Compared to purely figural tests such as the Raven
Matrices, this summed score represents to larger extent
knowledge. But conventional intelligence tests use not
only figural, but verbal and numerical tasks too, and
intelligence theory always subsumed under intelligence
the ability to process knowledge through the use of
reasoning skills and the understanding of complex
cognitive problems (Piaget, 1947; Cattell, 1987/1971;
Carroll, 1993).
4. Cognitive abilities and wealth: the problem of
causality
Lynn and Vanhanen (2002) have demonstrated a
strong correlation of r = .62 between intelligence test
results and GDP in 1998 (N = 185). Hanushek and Kimko
(2000), Jones and Schneider (2006), Ram (2007), Weede
and Kämpf (2002), Weede (2006) and Whetzel and
McDaniel (2006) showed that correlations of cognitive
ability with national wealth and economic growth persist
in regression models controlling for other plausible
macro-social factors and when using different variables
and measurement points. Positive correlations also exist
for state comparisons within the USA (r = .81, Davenport
& Remmers, 1950; r = .50 without Washington DC,
Kanazawa, 2006; r = .18–.28, McDaniel, 2006a,b). The
interpretation – cognitive ability increases economic
wealth by improving the quality of work in complex
129
situations – is based on the economic human capital
theory. A similar position is held by the economic
organization OECD and by many economic researchers
(e.g. Keating & Hertzman, 1999).
However, cross-sectional studies cannot present very
strong evidence for causal interpretations of the
cognitive-ability–GDP-relationship. Intelligence and
knowledge can influence GDP or vice versa or maybe
there is a reciprocal causality. As Ervik (2003)
maintained: “Many alternative explanations to the
causal interpretation of the results come to mind: for
example, richer countries spend much more on school-
ing and childcare, and schooling and IQ scores are
positively correlated. Hence, rich countries have higher
IQ scores.” (p. 407) We need “proof regarding the
direction of causality” (Barnett & Williams, 2004, p.
395). Because of these ambiguities, Whetzel and
McDaniel (2006) demand for the IQ–GDP-relationship
more “theory development and efforts at causal
modeling” (p. 457). Hunt and Wittmann (in press)
recommend longitudinal studies.
Longitudinal analyses that can distinguish the
influence of cognitive abilities or education on
national wealth from the influence of wealth on
education or cognitive abilities have not been done
so far. International data collections (Barro & Lee,
1993, 2000; Lee & Barro, 1997) allow such analyses.
Cross-lagged panel designs, which compare the
influence of competing factors in a defined historical
period, are the best feasible method for analyzing such
relationships. In these analyses the standardized path
coefficients between different variables are to be
interpreted, not the correlations (Rogosa, 1980). But
non-experimental studies and studies at the macro-
social level can never demonstrate in a compelling and
unequivocal way causal effects of variables free from
any historical context or free from the influence of
other variables:
1. Outcomes could be caused by unmeasured determi-
nants, which are correlated with intelligence, such as
economic freedom or political freedom. This could
lead both to over- and underestimation of the IQ
effect.
2. Other determinants could be important too, for
example natural resources. However, the aim of
this study is not to provide a comprehensive
explanation for national wealth and economic growth
(backward), but merely to study the relevance of
education and intelligence for these outcomes
(forward). Further the existence of other important
factors for wealth does not destroy the relevance of
130 H. Rindermann / Intelligence 36 (2008) 127–142
cognitive abilities to wealth. (For example: The
power of the sun to warm still exists on earth when
additional heat is produced by volcanoes.)
3. Researched determinants could be influenced by
other determinants and the effects of intelligence and
education can be indirect. For example, education
increases intelligence, intelligence increases the
quality of work and the efficiency of organizations
and institutions, which in turn increase wealth. But
factors behind or between (mediators) do not detract
from the importance of intelligence. They help to
understand more deeply the causal relationships.
4. Intelligence could have different effects in various
nations or historical periods. For example, intelli-
gence could be less important for national wealth in
countries at earlier stages of cultural, social and
economic development. Thus the results would
adequately describe the relationships for countries
in modernity.
The aim could only be to show the effect of cognitive
abilities controlling for important variables in a defined
historical period.
5. Aims of this study
The first aim of the study is to describe the
distribution of the new summary score of intellectual
ability (formed from IQ and school achievement scores)
across nations and its empirical validity (correlations
with measures of education, economic success and other
characteristics of societies).
The second aim is to investigate the influence of
education and cognitive abilities on GDP and vice versa
at the end of the 20th century with a cross-lagged
design, controlling for the usually most important
variable for economic growth: economic freedom.
6. Method
To reduce problems of poor data quality and to
ensure that countries at all levels of cultural, social and
economic development are represented in the sample, an
average score was formed from different measures of
one construct available for each country.
6.1. Cognitive abilities and corrections
The IQ-database of 113 countries stems from Lynn
and Vanhanen (2006). These IQs are standardized on a
common norm scale with the mean of Great Britain set
at 100 and the British standard deviation set at 15. Their
estimates for 79 states, based on measured data from
neighboring countries, were corrected down (if student
assessment studies are missing too) by 5 points. This
was done because it is assumed that countries with
missing data tend to be less developed with respect to
education and research, which in turn tends to be
associated with poorer cognitive development. For
Liechtenstein, which is missed completely in Lynn
and Vanhanen (but participated in PISA), the mean of
Switzerland, Austria and Germany (IQ 100) was used.
The correlations of uncorrected IQ with further
cognitive ability measures are: with PISA 2000 sum
uncorrected r = .88 (N = 46), with PISA 2000 sum
corrected r = .84 (N = 46), with TIMSS sum r = .89
(N = 63), with TIMSS sum corrected r = .88 (N = 63),
with PIRLS r = .78 (N = 33), with PIRLS corrected
r = .81 (N = 33), with GDP 1998 r = .62 (N = 185).
Correlations of corrected IQ including Liechtenstein:
with PISA 2000 sum uncorrected r = .88 (N = 47), with
PISA sum corrected r = .84 (N = 47), with TIMSS sum
r = .89 (N = 63), with TIMSS sum corrected r = .88
(N = 63), with PIRLS r = .78 (N = 33), with PIRLS
corrected r = .81 (N = 33), with GDP r = .63 (N = 185).
PISA-2000 (Programme for International Student
Assessment, OECD, 2003): The sum was formed from
the scores for reading, mathematics and science
literacy (41 states; α = .98). For PISA and other school
achievement studies (TIMSS, PIRLS) the mean was
set at 500 and the standard deviation at 100. Only 15-
year old students took part in PISA. Invariably, the
intellectual proficiency of all young people is over-
estimated for those countries in which the school
enrolment ratio of 15-year olds is low, for example
Albania (43%), Mexico (52%), Brazil (53%) and Peru
(66%). This is because schooling is one of the most
important determinants of cognitive development
(Lurija, 1976/1974; Ceci, 1991), and those no longer
in schools at age 15 are expected to have lower
intellectual proficiency than those still in schools.
Therefore the sum score was reduced according to the
following formula: PISA2000c = PISA2000 − ((100
− participation rate from 0 to 100) × 2). The correction
means: Countries with 90% attendance rate get 20 points
subtracted, with 80% attendance rate 40 points sub-
tracted. Other corrections are conceivable. The used
correction formula assumes a normal distribution from
which an end piece of the lower part has been deleted:
For example, when the lowest 16% of a normally
distributed sample with a mean IQ of 100 are missing,
the mean resulting IQ will be 104–105 (PISA: 530). The
used formula subtracts for IQ 4.8 IQ-points or for PISA
32 points from the measured average and leads to
 H. Rindermann / Intelligence 36 (2008) 127–142
marginally stronger corrections. The corrections are
stronger for extremely low school enrolment ratios, e.g.
for 50% missing the formula would subtract 15 IQ-points
or 100 PISA-points, although in a normally distributed
sample the resulting mean would be only about 13 IQ-
points or 87 PISA-points higher. The assumption is that
with low and extremely low school enrolment ratios
negative influences of other IQ-lowering factors, such as
poor nutrition and health care, low appreciation of
education, of reading books, of argumentation and
thinking in everyday life, and intellectual simplicity of
home environments and society, will increase. For a
complete list of correlations between variables see
Rindermann (in press).
PISA-2003 (OECD, 2004a, 2004b): The summary
score was formed for reading, mathematics, science and
problem solving (40 states; α = .99). This score was
corrected with the same formula as for PISA-2000
(PISA03K = PISA03 − ((100 − PartRP03) * 2)).
TIMSS 1994–1995 (Third International Mathematics
and Science Study, Beaton et al., 1996a; Beaton et al.,
1996b; Martin et al., 1997; Mullis et al., 1997), 8th
grade (N = 39) and 4th grade (N = 25). An averaged
score of mathematics and science literacy was formed
(α = .94 and α = .93), and this was corrected for mean
student age (formula for 8th grade: TI895k = TI895 −
((AgeTi958 − 14.308) * 42); 4th grade TI495k = TI495 −
((AgeTi954 − 10.232) * 42)). The correction means:
Countries with students 1 year older than the mean
(14.308 or 10.232 years) get 42 points subtracted, and
countries with students 1 year younger than the mean
get 42 points added. One year of school is expected to
represent a gain of ca. 42 points (mean of reports by
Mullis et al., 1997, p. 31 and p. 43, Martin et al., 1997,
p. 29 and p. 41, Beaton et al., 1996a, p. 29, Beaton et al.,
1996b, p. 29). For the 8th grade, information also exists
about participation rates (“coverage of 13-year old
students”; Beaton et al., 1996a, p. A12) and the
correction formula was: TI958cc = TI958c − (100 −
ParRT958) * 2. For Kuwait and Israel, no information
about attendance rates was provided, so the value of the
next Muslim neighbor Iran (72%) was used for Kuwait,
and the worldwide mean (87.57%) was used for Israel.
TIMSS 1999 (8th grade N = 37; Martin et al., 2000;
Mullis et al., 2000): The correction for older or younger
than mean aged students was: TI899k = TI899 −
(AgeTi998 − 14.366) * 42. The correction means: For
countries with students 1 year older than the mean
(14.366 years) 42 points were deducted, and for
countries with students 1 year younger than the mean
42 points were added. No information is presented in the
reports about participation rates.
131
TIMSS 2003 (8th grade N = 44, 4th grade N = 24;
Martin, Mullis, Gonzalez, & Chrostowski, 2004;
Mullis, Martin, Gonzalez, & Chrostowski, 2004). The
correction formulas for older or younger than mean
student age are for the 8th grade: TI803k = TI803 −
((AgeTi038 − 14.459) * 42); for the 4th grade: TI403 −
((AgeTi034 − 10.368) * 42). No information is presented
in the reports about participation rates.
IEA-Reading-Study 1991 (N = 24 countries for 9-year
olds and N = 31 countries for 14-year olds; Elley, 1992).
The mean age was 9.77 for 9-year olds and 14.73 years
for 14-year olds. The correction formula for older or
younger than mean student age and for low par-
ticipation rates of students are for the 9-year old student
study are: Read09k = Read09 − ((Age9 − 9.7692) * 42) −
((100 − PR9E) * 2); for the 14-year old student study:
Read14k = Read14 − ((Age14 − 14.7333) * 42) − ((100 −
PR14E) * 2).
PIRLS 2001 (Progress in International Reading
Literacy Study, N = 33 countries; Mullis, Martin,
Gonzalez, & Kennedy, 2003): Reading competences
of fourth graders. Correction for older or younger than
mean student age (formula: PIRLSK = PIRLS − ((Age
− 10.312) * 42)). The correction formula means: For
countries with students 1 year older than the mean
(10.312 years) 42 points were deducted and for
countries with students 1 year younger than the mean
were added 42 points. No information is presented in the
reports about participation rates.
Averaged score of all corrected measures of cognitive
abilities: Aggregation of different ability scales within
studies was done by taking the arithmetic mean. The
original scale was conserved. Between different studies,
different grades/ages of the same study in the same year
(e.g., in IEA-Reading 1991, and TIMSS 1995), different
years of the same study (PISA 2000 and 2003; TIMSS
1995, 1999 and 2003), and different studies (IEA-
Reading, TIMSS, PISA, PIRLS and IQ-collection),
aggregation was only possible after development of a
standardization formula with those countries that
participated in at least two studies. This was required
because different scales had different means and
standard deviations (e.g., 500 and 100 vs. 100 and
15), and different assessments had different standardiza-
tion samples (but the same scales, e.g., IEA-Reading,
TIMSS and PISA, or TIMSS 1995 grade 4 vs. 8; PISA
2000 vs. 2003). Before aggregation, the means and
standard deviations for those countries that took part in
two (or more) studies were calculated. The aggregation
was done stepwise: First, the different scales of one
study were averaged (e.g., for TIMSS 1999: mathe-
matics and science); then different age or grade levels of
132 H. Rindermann / Intelligence 36 (2008) 127–142
one study in 1 year (e.g. IEA-Reading 1991 9- and
14 years old, TIMSS 1995 4th and 8th grade); then
different survey years of one study type (PISA 2000 and
2003, or TIMSS 1995, 1999 and 2003); and finally
different studies of one student assessment approach
(age, PISA-studies, vs. class, TIMSS-studies and
PIRLS, vs. mixed and old study, IEA-Reading). At the
end, one total score was calculated for all cognitive
ability studies. Student assessment studies were given
twice the weight of the Lynn and Vanhanen (2006) IQ
collection because they are more recent, have larger
samples, and consist of more studies. A similar total
score was calculated for student assessment studies only
(PISA, TIMSS, PIRLS, IEA-Reading).
Correlations of both sum scores, first total sum value,
second student assessment sum value, with single study
measures are: total sum value with IQ-L&V corrected
r = .99 (N = 193) and student assessment value with IQ-
L&V corrected r = .86 (N = 77), with PISA sum
corrected r = .97 (N = 48) and r = .99 (N = 48), TIMSS
sum corrected r = .97 (N = 63) and r = .98 (N = 63), IEA-
Reading corrected r = .95 (N = 31) and r = .96 (N = 31),
PIRLS corrected r = .95 (N = 33) and r = .97 (N = 33).
Both sum scores (all cognitive ability studies and
student assessment studies only) correlate with r = .98
(N = 78).
The results of the different cognitive ability studies
are highly correlated. Outcomes of factor analyses
support an unidimensional factor structure (93–95% of
the variance was explained by the first unrotated factor;
Rindermann, 2006, in press).
6.2. Attributes of societies
Educational level of adults: The standardized values
of three measures were averaged (α = .94, N = 173
countries): 1. Adult literacy rate 1991 (N = 172), from
Vanhanen (1997). 2. Percentage of persons between 12
and 19 years old 1960–1985 (today adults) having
graduated from secondary school (N = 117), from
Mankiw, Romer, and Weil (1992). 3. The mean of
years of schooling of persons 25 years old or older for
1990, 1995 and 2000 (N = 107), from Barro and Lee
(2000).
Quality and quantity of school education of students:
Standardized scores were averaged for the percentage of
children attending pre-school or kindergarten, hours
taught per year in regular schools, young age of
streaming (as an indicator of emphasis on school per-
formance, streaming in younger age leads to a higher
value), use of performance tests by schools for coun-
seling parents or streaming decisions, central exams,
discipline (low skipping rate, punctuality, low rate of
interruptions or behavioral problems during teaching),
supplementary lessons during afternoon, evening, week-
end or holidays, small classes and low student–teacher-
ratios, and reaching of high grades at a young age (e.g.,
in the age of 10 pupils at grade 5 and not in the age of 11
at grade 4). Data are collected from the OECD- and
IEA-studies and from Barro and Lee (1993), α = .59,
N = 158.
Gross domestic product (purchasing power parity)
per capita 1998 from Lynn and Vanhanen (2002),
N = 185. Their sources are UN data sets.
Rule of law 1970–2000 with emphasis on ownership
law from Gwartney and Lawson (2003) and Knack and
Keefer (1995), α = .90, N = 131.
Speed of life 1992–1995 from Levine (1997),
containing service speed at post office (how much
time is needed to sell a stamp), walking speed and
accuracy of clocks (α = .72, N = 31).
Number of books (parents and children) from PIRLS
2001 and TIMSS 1995, 1999 and 2003 (α = .94, N = 63).
Democracy 1950–2004 from Vanhanen (2005) and
Marshall and Jaggers (2000), α = .95, N = 183.
Interpersonal trust in 1990s from Inglehart (1997),
N = 41.
Economic growth between 1950 and 1990 from Lynn
and Vanhanen (2002), N = 185. Their source is Maddi-
son (1995).
Homicide rate (per 100 000 inhabitants) 1995 and
2002 (α = .41, N = 138) and rate of solved cases for
homicide 1995 and 2002 (α = .77, N = 122) from Interpol
(2004).
War 1960–1990, N = 118, from Sala-i-Martin (1997)
completed and corrected by the author for the years
between 1960 and 2000, N = 186. The variable
represents not only participation in war, but intensity
of war and the destructive effect of war in their own
country including civil war (e.g. USA small value in
spite of participation in many wars; Afghanistan, Haiti,
Liberia and Somalia maximum value). Correlation
between war from Sala-i-Martin and the new war
variable: r = .66.
HIV-infection rate 2001 and 2003 for adults from
UNAIDS/WHO (2003), α = .99, N = 165.
Government spending ratio 1980–89 from Barro and
Lee (1993), N = 138. Percentage of government spend-
ing (e.g. for redistribution, for administration, for police,
for military, for infrastructure) of gross domestic product.
Gini-coefficient (income inequality) from Fukuda-
Parr (2004) and Deininger and Squire (1996), 1951–
2002 period averaged (major period 1970s to 1990s),
α = .89, N = 148.
 H. Rindermann / Intelligence 36 (2008) 127–142
Number of children per woman (total fertility rate,
TFR), 1960–84 average from Barro and Lee (1993),
N = 130.
Economic freedom 1960–2000 from Gwartney and
Lawson (2003), O'Driscoll, Holmes, and O'Grady
(2002) and Sala-i-Martin (1997), α = .89, N = 165.
Level of low corruption 1980–2003 (corruption
inverted to low corruption) from Transparency Interna-
tional (2004), α = .98, N = 132.
Quality and speed of bureaucracy between 1972 and
1995 from Knack and Keefer (1995), N = 60, and
between 1984 and 2002 from Drori, Jang, and Meyer
(2006), N = 140, α = .90, N = 140. Both author groups
use data from international organizations (Business
Environmental Risk Intelligence, International Country
Risk Group).
Geographic distance from equator from Sala-i-
Martin (1997), N = 133.
6.3. Data for cross-lagged study (repeated
measurements)
Years at school was the “average schooling years
in the total population over age 25” according to Barro
and Lee (2000). (For 1970 N = 101 countries, for 2000
N = 104 countries.)
GDP was from Barro and Lee (1993) for 1970 (122
countries), and from Lynn and Vanhanen (2002) for
1998 (185 countries).
Economic freedom ratings for 1970 and 2000 (122
countries each) were from the Fraser Institute (Gwartney
& Lawson, 2003).
Measures from former studies on cognitive abilities
come from student assessment studies collected by Lee
and Barro (1997). From 1964: IEA-Mathematics 13-year
old students, eighth grade; IEA-Mathematics at the end
of secondary school. From 1972: Science 10-year old
students; science 14-year old students; science at the end
of secondary school; reading 13-year old students. Mean
correlation with weighted N and after Fisher's-Z-
transformation is r = .62. The complete sample for old
student assessment studies includes 19 nations: Aus-
tralia, Belgium, Chile, Finland, France, Germany, Great
Britain, Hungary, India, Iran, Israel, Italy, Japan, Malawi,
Netherlands, New Zealand, Sweden, Thailand, USA.
The overall cognitive ability score including intelli-
gence tests from Lynn and Vanhanen is not usable as a
second measurement point of cognitive abilities because
in many countries part of the IQ values is of older origin.
Therefore only data from student assessment studies
between 1991 and 2003 were used for the cross-lagged
analysis.
133
The time period match is not perfect, but it
approximates a 30-year interval. The use of GDP from
1998 and 2000 leads to nearly identical results. The
1998 data show one advantage: 1998 is nearly half-way
between 1991 and 2003, the time range of the more
recent set of student assessment studies.
6.4. Cognitive ability data only from the youth of
nations — a problem?
The use of large-scale assessments from students
between 9 and 15 years of age to explain macro-social
development (e.g. growth) seems to be questionable.
Students do not work, vote or make political decisions.
However, intelligence is fairly stable during adult life,
and therefore adolescents represent a plausible proxy for
the intelligence level of the working population during
the following decades. In addition, the school achieve-
ment of children and adolescents closely parallels the
intellectual ability level of adults in the country. This is
shown in a correlation of r = .70 between the average
scores of student performance studies and the results of
the adult literacy study of the OECD (2000; N = 20,
r = .73 with the total score of all cognitive ability studies;
Rindermann, in press). Therefore differences between
nations in the adolescent population correlate with
differences between nations in the adult population.
Differences in reproduction rates of the adults could
cause problems in cross-sectional, but not in long-
itudinal studies.
6.5. Statistical methods
Pearson product moment correlations were used for
bivariate correlations, additionally GDP was partialed
out. Longitudinal effects were calculated by the use of
cross-lagged path coefficients in a cross-lagged panel
design (see Shadish, Cook, & Campbell, 2002; for
causal interpretation: Pearl, 2000). This method pro-
vides a test of reciprocal causal relations between two or
more variables. The standardized path coefficients (β)
between time-lagged variables are reported, along with
correlations in parentheses. Additional correlations help
to estimate the influence of other variables in the model
(by inspection of the difference between the correlation
coefficient and the path coefficient), they allow a check
of the model (1 − error = R2 = Σrβ) and to calculate the
proportion of explained variance through each factor
(R2 = Σrβ). According to Rogosa (1980), unlike the path
coefficients the cross-lagged correlations are not useful
for estimating causal effects because of their stronger
dependence on the stability and variance of the
134 H. Rindermann / Intelligence 36 (2008) 127–142
variables. An even more important reason is that cross-
lagged path coefficients represent the incremental part
of the other variables in the model, the part that is not
explained by self-prediction. Even highly stable vari-
ables, such as GDP, can be explained by other variables
in a model. The cross-lagged path analyses were done
with LISREL. Good values for fit indices are
SRMR ≤ .08 (Hu & Bentler, 1999) or SRMR ≤ .05
(Schermelleh-Engel, Moosbrugger, & Müller, 2003) and
CFI ≥ .95 (Hu & Bentler, 1999) or CFI ≥ .97 (Scher-
melleh-Engel et al., 2003).
Cross-lagged panel designs can explain the changes
in a variable from a former to a subsequent measurement
point. They cannot explain the origin of differences at
the first measurement point. Generalization to other time
periods is possible only with additional assumptions,
justification and empirical evidence (e.g. “because of the
relevance of invention, organization and planning,
intelligence and knowledge in developed societies
have been important for growth from 1950 to 1970 as
well”). For longitudinal cross-lagged panel designs GDP
is a more appropriate variable than economic growth
(the difference between two measurement points itself
represents growth, the change in wealth).
Like all non-experimental designs, cross-lagged
panel designs are useful for causal interpretations only
if the “right” variables are included. The non-inclusion
of causal variables that are correlated with the variables
in the analysis can bias the results. A less relevant
problem is the absence of background variables like
culture, because they probably stand behind the
researched variables and mainly work through them
(e.g. culture influences education, cognitive abilities and
economic freedom and through these factors culture has
influence on national wealth). To reduce problems of
causal interpretation, the conventionally most important
variable for economic growth, economic freedom, was
included, different samples were used (for educational
variables there exists a larger sample) and in further
studies political aspects were considered (Rindermann,
submitted for publication). Additional variables, such as
mineral resources, are important for national wealth
(Lynn & Vanhanen, 2002, 2006; Whetzel & McDaniel,
2006), but they do not change the influence of
education, cognitive abilities and economic freedom.
Significance tests were not used for interpretation
(for an in-depth justification e.g. Cohen, 1994; Falk &
Greenbaum, 1995; Gigerenzer, 2004; Hunter, 1997).
Especially at the macro-social level they are not
appropriate for scientific reasoning. More instructive
for inductive generalization – which is not possible with
significance tests – is the demonstration of the stability
of relationships across different country samples,
different variables, different measurement points and
various studies of different authors.
For drawing the map the SAS program was used.
7. Results
7.1. International distribution of intelligence and
knowledge
Mean cognitive abilities differ greatly between
countries (Fig. 1). The East Asia region has the highest
cognitive ability scores, including Taiwan (first),
Singapore (second), China (third), South Korea (fourth),
Hong Kong (fifth) and Japan (sixth). Second are North-,
West- and Central-Europe and the countries which have
been settled by these Europeans: Finland (seventh),
Netherlands (ninth), Canada (tenth), Great Britain
(eleventh), Iceland (twelfth), Switzerland (13th), Austria
(15th), New Zealand (16th), Sweden (17th), Australia
(18th), Belgium (20th), Norway (21st), USA (23rd),
France (24th), Germany (25th) and Denmark (27th).
Medium results are achieved for South and East Europe
(here the highest results are for: Italy with 14th and
Czech Republic with 19th rank) and low results for
Latin America (here the highest rank is for Uruguay
with 55th).
Scores are low in the Muslim Middle East and even
lower in sub-Saharan Africa. With one exception, all of
the 34 lowest-scoring countries are in sub-Saharan
Africa. The exception is Haiti (185th, data were
estimated by Lynn and Vanhanen by using test results
from Jamaica, estimated intelligence test data were
corrected down by 5 points), a country with a population
and culture similar to sub-Saharan Africa. Even some
countries with good economic development and rela-
tively stable democracy and rule of law, for example
Botswana (138th) and South Africa (155th), score low
on measures of cognitive ability. Intervention studies
using formal education or training have demonstrated an
unused potential for cognitive improvement in these
countries (Skuy et al., 2002). Extremely rich Arab
countries such as Kuwait (100th) score no better than
poor Arab countries (Lebanon, 93rd; Iran, 81st).
Regions with low cognitive abilities, including sub-
Saharan Africa and the Muslim Middle East, seem to be
prone to cultural and social crisis (see UNDP, 2003;
UNESCO, 2004). Low political power and political
instability are associated with low cognitive ability.
These observations suggest important causal rela-
tionships of cognitive abilities with political, cultural
and economic conditions. But are the cognitive ability
 H. Rindermann / Intelligence 36 (2008) 127–142
Fig. 1. World map of cognitive ability sum score (N = 194, darker means higher ability, no data for West Sahara and Greenland).
test measurements and the national differences valid?
And if they are valid, what are the causes of national
differences in cognitive abilities?
7.2. Correlations with important macro-social
variables
Correlations (Pearson's r) with country-level vari-
ables, mainly from the 1990s, show close relations
between cognitive abilities and educational attributes
(Fig. 2). In countries with young adults of a high
educational level, there are good quality and quantity of
school education and plenty of books at home, with the
result that intelligence test and school performance test
scores are higher.
But there are also high correlations with social,
political and economic variables including quality and
speed of bureaucracy, rule of law, gross domestic product
(GDP), low corruption, democracy, economic freedom
and economic growth. The correlation between cogni-
tive ability level and GDP (r = .63, N = 185) even
increases when the logarithm of GDP is used (r = .70,
N = 185), but remains below the correlations with
education (educational level of adults r = .78, N = 173,
quality and quantity of school education r = .74,
N = 158). An increase of 10 IQ points corresponds to a
doubling of the GDP (Dickerson, 2006). Even inter-
personal trust and the rate of solved homicide cases are
associated positively with intelligence and knowledge.
Negatively correlated are homicide rates, war (destruc-
tion by war; see also DeGroot, 1951), HIV-infection
rates, high government spending ratio, income inequal-
ity and the child rate per woman (total fertility rate). The
135
negative correlation with homicide corresponds to
similar results at the individual and US-state level
(Lynn & Vanhanen, 2002; McDaniel, 2006b), the
negative correlation with HIV is mirrored by a positive
correlation with health at US-state level (McDaniel,
2006b).
Even after partialing out GDP, educational variables
(education of young adults, education of the youth and
number of books) remain highly correlated with
cognitive abilities. Political, economic and social
attributes of societies show less stable correlations.
The positive correlation between adult education and
cognitive ability remains stable in different international
sub-groups: in developed r = .56 (N = 82) or less
developed countries r = .55 (N = 91); in sub-Saharan
Africa r = .48 (N = 46) or in Europe r = .45 (N = 37); in
traditionally Catholic countries r = .62 (N = 36), Protes-
tant countries r = .21 (N = 13), Orthodox countries
r = .95 (N = 13), countries of eastern religions (Con-
fucian, Buddhist, Shinto) r = .73 (N = 15) and Muslim
countries r = .71 (N = 32). After controlling for distance
from the equator, the partial correlation between adult
education and cognitive ability is rp = .63 (N = 133).
The ability to think and to use knowledge seems to be
attributable to education (see Barber, 2005). Differences
in cognitive abilities at international level reflect
differences in education. This has been shown in
many studies at the individual level (Lurija, 1976/
1974; Ceci, 1991; Stelzl, Merz, Remer, & Ehlers, 1995;
Winship & Korenman, 1997). The same is demonstrated
here at the country level, with high and stable
correlations between cognitive measures and school
education.
136 H. Rindermann / Intelligence 36 (2008) 127–142
Fig. 2. Correlations between cognitive ability and society attributes.
The correlational pattern shows that intelligence and
knowledge are indicators, causes and/or consequences
of a successful civil society (see Rindermann, submitted
for publication). One causal hypothesis will be tested
here: the economic wealth thesis.
7.3. Cross-lagged relationships between education and
GDP and between cognitive ability and GDP
Lynn and Vanhanen (2002) have demonstrated a
correlation of r = .62 between IQ and GDP98 at the
country level (N = 185). After downward correction of
their estimates for countries with missing data this
correlation is r = .63. The correlation of PISA 2000 with
GDP98 is even higher despite substantial range
restriction (r = .69, N = 40; r = .72 with correction for
participation rate). Other authors have confirmed the
results. In direct comparisons, GDP is more strongly
associated with cognitive ability measures (r = .65,
N = 185) than with educational variables (r = .60,
N = 173). The difference is even more marked for the
regression coefficients in a 2-predictor model
(βCA → GDP = .48, r = .65, βEdu → GDP = .22, r = .60,
N = 173). Educational information is less reliable and
valid than are ability measures. And years in school and
academic degrees (e.g. “secondary school complete”)
are less important than the quality of the product that
leaves the schools. But cross-sectional studies do not
reveal the direction of the causal arrow between
cognitive ability and GDP. Intelligence and knowledge
can influence GDP (human capital theory) or vice versa
(e.g. by better nutrition and health care) or maybe there
is a reciprocal causation. This question can best be
approached through the use of longitudinal cross-lagged
analysis.
For the path analysis in Fig. 4 (years of schooling) the
fit indices are good: SRMR = .02 and CFI = .97. Educa-
tion has a stronger impact than economic freedom
(including aspects of rule of law) on GDP (βSY1 → GDP2 =
.40, r = .84, vs. βEF1 → GDP2 = .23, r = .68; see also Table
1). The longer and the more the children of 1970
attended school, the better was the economic develop-
ment between 1970 and the end of the century.
This analysis included two ex-communist countries,
Hungary and Poland. When these two countries are
excluded, the results for the remaining 86 countries
remain the same: The impact of years in school (SY) on
GDP (βSY1 → GDP2 = .43, r = .86) is stronger than the
effect of economic freedom on GDP (βEF1 → GDP2 = .22,
r = .69), and even stronger than the effect of old GDP on
 H. Rindermann / Intelligence 36 (2008) 127–142 137

Table 1 Here we are limited to a sample of only 17 nations

Coefficients for cross-lagged analyses that includes Australia, Belgium, Chile, Finland, France,
Relationship r (p) β (p) CI of β Germany, Great Britain, Hungary, Iran, Israel, Italy,
Fig. 4, education, economic freedom and gross domestic product, Japan, Netherlands, New Zealand, Sweden, Thailand,
N = 88, 1970–1998/2000 and the USA. But the results are virtually the same (Fig.
SY1 → SY2 .93 (.000) .93 (.000) .77–1.07 5). For the path analysis in Fig. 5 the fit indices are very
SY1 → EF2 .71 (.000) .54 (.000) .27–.76
good: SRMR = .02 and CFI = 1.00. Again, economic
SY1 → GDP2 .84 (.000) .40 (.000) .27–.63
EF1 → SY2 .51 (.000) .09 (.084) −.01–.19 growth is more strongly related to school-related
EF1 → EF2 .76 (.000) .31 (.001) .13–.48 cognitive abilities (βCA1 → GDP2 = .29) than to economic
EF1 → GDP2 .68 (.000) .23 (.000) .14–.39 freedom (βEF1 → GDP2 = .10).
GDP1 → SY2 .78 (.000) −.06 (.468) −.22–.10 When ex-communist Hungary is excluded from the
GDP1 → EF2 .67 (.000) .02 (.891) −.25–.29
sample, the results remain similar. The impact of
GDP1 → GDP2 .88 (.000) .40 (.000) .23–.63
cognitive abilities on GDP (βCA1 → GDP2 = .34, r = .73)
Fig. 5, cognitive ability, economic freedom and gross domestic is still stronger than the impact of economic freedom
product, N = 17, 1964/72–1991/2003 (βEF1 → GDP2 = .08, r = .51).
CA1 → CA2 .85 (.000) .74 (.000) .37–.84 The results remain similar when the logarithm of
CA1 → EF2 .55 (.024) .25 (.268) .02–.79
GDP is used for the 17-nations sample, with a stronger
CA1 → GDP2 .63 (.007) .29 (.024) .05–.54
EF1 → CA2 .38 (.129) .17 (.263) −.08–.31 impact of cognitive abilities (βCA1 → GDP2 = .38, r = .72)
EF1 → EF2 .25 (.332) −.20 (.437) −.54–.27 than economic freedom (βEF1 → GDP2 = .03, r = .45), and
EF1 → GDP2 .57 (.017) .10 (.435) −.18–.34 a stronger impact of cognitive abilities on gross do-
GDP1 → CA2 .66 (.004) .21 (.237) −.08–.28 mestic product than gross domestic product on cognitive
GDP1 → EF2 .68 (.003) .68 (.028) .07–.91
abilities (βGDP1 → CA2 = .20, r = .67). Thus education and
GDP1 → GDP2 .90 (.000) .70 (.000) .42–.96
cognitive abilities are more important than economic
Note: standardized path coefficients (β), exact significance levels up to
freedom as determinants of economic growth.
three decimal places, CI: confidence interval (95%) for standardized
path coefficients; SY: years of school education, EF: economic Differences between nations in education and
freedom, GDP: gross domestic product, CA: cognitive ability. cognitive abilities explain better later differences in
wealth than differences in economic freedom do.
But national wealth influences the development of
new GDP (βGDP1 → GDP2 = .37, r = .88). Reverse effects cognitive abilities too (βGDP1 → CA2 = .21; see Fig. 5).
on education (years at school) are small: βEF1 → SY2 = .09 Good nutrition and health care are considered important
(r = .55), βGDP1 → SY2 = −.08 (r = .79) and βSY1 → SY2 = for the development of cognitive abilities, especially
.95 (r = .93). among the poor sections of society (Lynn, 1990;
By using the logarithm of GDP the results remain Glewwe & King, 2001; Whaley et al., 2003), probably
similar for the entire 88-nations sample, with a stronger also the access to mass media like newspaper, TV and
impact of education (βSY1 → GDP2 = .27, r = .84) than Internet (Barber, 2006). These factors are likely to be
economic freedom (βEF1 → GDP2 = .10, r = .60) on GDP. important for cognitive differences among countries as
Effects on education are small: βEF1 → SY2 = .03 (r = .51) well.
and βGDP1 → SY2 = .18 (r = .85). The correlations between macro-social variables
Formal education can create economic wealth only increase from 1970 to 2000. They may reflect the
by increasing human capital. It does so not by conferring growing interdependence of the success of nations in
educational degrees, but by changing the attitudes and modern age.
raising the cognitive ability of students. In the presented
analyses, education is treated as a proxy for cognitive 8. Discussion
ability, the capacity to understand and solve problems by
thinking and using of knowledge. But the effects of The high correlations with a host of macro-social
education can go beyond knowledge and reasoning variables (Figs. 2 and 3) demonstrate the validity of
skills (including attitude change such as conscientious- intelligence as a link in the causal nexus of cultural,
ness and cooperativeness). In order to confirm that the political and economic “modernization”. Measures of
effects of schooling are mediated through increased formal schooling, in particular, are closely related to
cognitive skills, we have to repeat the analysis with measures of cognitive ability. International differences
cognitive ability measures substituting for the measure in cognitive abilities correlate with differences in
of schooling. educational levels. These correlations especially support
138 H. Rindermann / Intelligence 36 (2008) 127–142
Fig. 3. Partial correlations between cognitive ability and society attributes (GDP partialed out).
the validity of international intelligence comparisons.
Education itself probably depends mainly on cultural
factors, less on economic. The education–intelligence
relationship is presumably reciprocal: schooling raises
intelligence, and intelligent people realize the advan-
tages to be gained through a better education. This does
not negate the importance of other factors for cognitive
development. Prosperity (Fig. 5) and genes (Rushton &
Jensen, 2005) are all likely to contribute also to
intelligence differences between nations and groups.
The economic, social and cultural complexity of
modern societies would be impossible without a high
level of cognitive abilities. Conversely, high intelli-
gence would not develop without this complexity. The
20th century has seen near-exponential advances in
science and technology, massive increases in the
complexity of commercial enterprises and government
bureaucracies and, perhaps most important, a vast
expansion of the school system (Meyer, Ramirez, &
Soysal, 1992; Schofer & Meyer, 2005). This has been
accompanied by large IQ gains, which have been in
excess of 3 points per decade in many countries during
much of the 20th century (Flynn, 1987). Thus cognitive
development and cultural modernization are intimately
associated (Oesterdiekhoff, 2000). It has been proposed
that the unusual developments of the past one or two
centuries are based on positive feedback loops in which
increased and better school education, technological
advances and rising prosperity raised intelligence, and
rising intelligence entailed further advances in technol-
ogy, prosperity and the school system (Meisenberg,
Lawless, Lambert, & Newton, 2005; Meisenberg, in
press).
The results reported here show that during the last
third of the 20th century, education and cognitive
abilities were more important for economic wealth
than economic wealth was for education and cognitive
abilities. This result is stable across the different
national samples of education and ability and remains
after adding additional factors like economic freedom.
Intelligence is even more important for wealth than
economic freedom (see also Weede, 2006)! Whereas
the importance of intelligence for many personal life
outcomes has been recognized for some time (Gott-
fredson, 2003; Herrnstein & Murray, 1994), we should
realize that intelligence is also an important determi-
nant for the economic and social development of
nations (for example the functioning of institutions in
the systems of law, economics and politics). The
present study shows that a high level of cognitive
 H. Rindermann / Intelligence 36 (2008) 127–142 139

Fig. 4. Standardized path coefficients (and correlations in parentheses) between average schooling years in the total population over age 25, economic
freedom and gross domestic product (error terms as unexplained variance on the right), N = 88 nations, only measured data.

development can be an antecedent and likely cause for
economic growth, but other macro-social outcomes
(e.g., democracy, rule of law, national power or health)
are likely to be influenced by education and intelli-
gence as well (Rindermann, submitted for publication;
Rindermann & Meisenberg, submitted for publication).
Certainly the positive influence of young people's
schooling and intelligence on the level of economic
freedom 30 years later (Figs. 4 and 5) deserves further
investigation. Future theoretical and empirical research
has to analyze the causal mechanism underlying the
effects of ability on development of societies in a more
detailed manner. For example, there is a positive
relationship with low government spending ratio
(r = .47 and rp = .24). Abilities seem to enable a more
liberal economic constitution and thriftiness of state
interventions. Conversely, a population with low
education and intelligence seems to necessitate more
state intervention, which tends to widen the influence
of powerful special-interest groups.

Fig. 5. Standardized path coefficients (and correlations in parentheses) between cognitive abilities (students' assessment studies from 1964 to 1972
and 1991 to 2003), economic freedom and gross domestic product (error terms as unexplained variance on the right), N = 17, only measured data.
140 H. Rindermann / Intelligence 36 (2008) 127–142
Acknowledgements
I have benefited from a lot of suggestions, help and
critics received after talks, lectures and distribution of
preliminary drafts of this paper by colleagues (especially
by Gerhard Meisenberg, Georg W. Oesterdiekhoff, and
Erich Weede) and students from Germany, Austria and
the USA and by the three anonymous reviewers.
References
Armor, D. J. (2003). Maximizing intelligence. New Brunswick:
Transaction Publishers.
Barber, N. (2005). Educational and ecological correlates of IQ: A
cross-national investigation. Intelligence, 33, 273−284.
Barber, N. (2006). Is the effect of national wealth on academic
achievement mediated by mass media and computers? Cross-
Cultural Research, 40, 130−151.
Barnett, W. S. (1998). Long-term effects on cognitive development and
school success. In S. W. Barnett, & S. S. Boocock (Eds.), Early care
and education for children in poverty. Promises, programs, and
long-term results (pp. 11−44). New York: State University.
Barnett, S. M., & Williams, W. (2004). National intelligence and the
emperor's new clothes. Contemporary Psychology, 49, 389−396.
Barro, R. J., & Lee, J.-W. (1993). Barro-Lee data set. International
measures of schooling years and schooling quality. Distributed by
the World Bank Group. Washington, Version 1994. Retrieved July
14, 2004 from www.worldbank.org.research/growth/ddbarle2.htm
Barro, R. J., & Lee, J.-W. (2000). Barro-Lee data set. International
data on educational attainment: Updates and implications.
Boston: Harvard University. Retrieved November 18, 2004 from
www2.cid.harvard.edu/ciddata/barrolee/readme.htm
Beaton, A. E., Mullis, I. V. S., Martin, M. O., Gonzalez, E. J., Kelly,
D.L., & Smith, T. (1996a). Mathematics achievement in the middle
school years. Chestnut Hill: TIMSS Study Center.
Beaton, A. E., Martin, M. O., Mullis, I. V. S., Gonzalez, E. J., Smith,
T., & Kelly, D. L. (1996b). Science achievement in the middle
school years. Chestnut Hill: TIMSS Study Center.
Carroll, J. B. (1993). Human cognitive abilities. A survey of factor-
analytic studies. Cambridge: Cambridge University Press.
Cattell, R. B. (1987/1971). Intelligence: Its structure, growth and
action. Amsterdam: Elsevier.
Ceci, S. J. (1991). How much does schooling influence general
intelligence and its cognitive components? A reassessment of the
evidence. Developmental Psychology, 27, 703−722.
Cohen, J. (1994). The earth is round (p < .05). American Psychologist,
49, 997−1003.
Coleman, W. C., & Cureton, E. E. (1954). Intelligence and
achievement: The ‘jangle fallacy’ again. Educational and
Psychological Measurement, 14, 347−351.
Cronbach, L. J. (1984). Elements of psychological testing. New York:
Harper and Row.
Davenport, K. S., & Remmers, H. H. (1950). Factors in state
characteristics related to average A-12 V-12 test scores. Journal of
Educational Psychology, 41, 110−115.
Deary, I. J., Strand, S., Smith, P., & Fernandes, C. (2007). Intelligence
and educational achievement. Intelligence, 35, 13−21.
DeGroot, A. D. (1951). War and the intelligence of youth. Journal of
Abnormal and Social Psychology, 46, 596−597.
Deininger, K., & Squire, L. (1996). Measuring income inequality: A
new database. New York. Retrieved September 20, 2004 from
www.worldbank.org/research/growth/dddeisqu.htm
Dickerson, R. E. (2006). Exponential correlation of IQ and the wealth
of nations. Intelligence, 34, 291−295.
Drori, G. S., Jang, Y. S., & Meyer, J. (2006). Sources of rationalized
governance: Cross-national longitudinal analyses. Administrative
Science Quarterly, 51, 205−229.
Elley, W. B. (1992). How in the world do students read? The Hague:
The International Association for the Evaluation of Educational
Achievement (IEA).
Ervik, A. O. (2003). IQ and the wealth of nations. Economic Journal,
113, 406−408.
Falk, R., & Greenbaum, Ch. W. (1995). Significance tests die hard.
The amazing persistence of a probabilistic misconception. Theory
and Psychology, 5, 75−98.
Flynn, J. R. (1987). Massive IQ gains in 14 nations: What IQ tests
really measure. Psychological Bulletin, 101, 171−191.
Frey, M. C., & Detterman, D. K. (2004). Scholastic assessment or g?
The relationship between the Scholastic Assessment Test and
general cognitive ability. Psychological Science, 15, 373−378.
Fukuda-Parr, S. (2004). Human development report 2004. New York.
Retrieved September 20, 2004 from http://hdr.undp.org/statistics/
data/indic/indic_138_1_1.html
Gigerenzer, G. (2004). Mindless statistics. Journal of Socio-Econo-
mics, 33, 587−606.
Glewwe, P., & King, E. (2001). The impact of early childhood
nutritional status on cognitive development: Does the timing of
malnutrition matter? World Bank Economic Review, 15, 81−113.
Gottfredson, L. (2003). g, jobs and life. In H. Nyborg (Ed.), The
scientific study of general intelligence. Tribute to Arthur R. Jensen
(pp. 293−342). Oxford: Pergamon.
Gottfredson, L., & Deary, I. J. (2004). Intelligence predicts health and
longevity, but why? Current Directions in Psychological Science,
13, 1−4.
Greenfield, P. M. (1997). You can't take it with you: Why ability assess-
ments don't cross cultures. American Psychologist, 52, 1115−1124.
Gwartney, J., & Lawson, R. (2003). Economic freedom of the world.
2003 annual report. Vancouver: Fraser Institute.
Habermas, J. (1976). Geschichte und Evolution. Geschichte und
Gesellschaft, 2, 310−357.
Hanushek, E. A., & Kimko, D. D. (2000). Schooling, labor-force
quality, and the growth of nations. American Economic Review,
90, 1184−1208.
Herrnstein, R. J., & Murray, Ch. (1994). The bell curve. Intelligence
and class structure in American life. New York: Free Press.
Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in
covariance structure analysis: Conventional criteria versus new
alternatives. Structural Equation Modeling, 6, 1−55.
Hunt, E., & Sternberg, R. J. (2006). Sorry, wrong numbers: An
analysis of a study of a correlation between skin color and IQ.
Intelligence, 34, 131−137.
Hunt, E. & Wittmann, W. (in press). National intelligence and national
prosperity. Intelligence, 35.
Hunter, J. E. (1997). Needed: A ban on the significance test.
Psychological Science, 8, 3−7.
Inglehart, R. (1997). Modernization and postmodernization. Cultural,
economic, and political change in 43 societies. Princeton:
Princeton University Press.
Interpol (2004). International Crime Statistics. Lyon. Retrieved
October 22, 2004 from www.interpol.int/Public/Statistics/ICS/
downloadList.asp
 H. Rindermann / Intelligence 36 (2008) 127–142
Jensen, A. R. (1998). The g factor. The science of mental ability.
Westport: Praeger.
Jones, G., & Schneider, W. J. (2006). Intelligence, human capital, and
economic growth: A Bayesian Averaging of Classical Estimates
(BACE) approach. Journal of Economic Growth, 11, 71−93.
Kanazawa, S. (2006). IQ and the wealth of states. Intelligence, 34,
593−600.
Keating, D., & Hertzman, C. (Eds.). (1999). Development health and
the wealth of nations New York: Guilford Press.
Knack, S., & Keefer, Ph. (1995). Institutions and economic
performance: Cross-country tests using alternative institutional
measures. Economics and Politics, 7, 207−227.
Lee, J.-W., & Barro, R. J. (1997). Schooling quality in a cross-section
of countries. New York. Retrieved November 18, 2004 from www.
nber.org/papers/W6198
Levine, R. (1997). A geography of time. New York: Basic Books.
Lurija, A. R. [Luria] (1976/1974). Cognitive development. Its cultural
and social foundations. Cambridge: Harvard University Press.
Lynn, R. (1990). The role of nutrition in secular increases in
intelligence. Personality and Individual Differences, 11, 273−285.
Lynn, R., & Vanhanen, T. (2002). IQ and the wealth of nations.
Westport: Praeger.
Lynn, R., & Vanhanen, T. (2006). IQ and global inequality. Athens:
Washington Summit.
Maddison, A. (1995). Monitoring the world economy, 1820–1992.
Paris: OECD.
Mankiw, N. G., Romer, D., & Weil, D. N. (1992). A contribution to the
empirics of economic growth. Quarterly Journal of Economics,
107, 407−437.
Marshall, M. G., & Jaggers, K. (2000). Polity IV Project. Political
regime characteristics and transitions, 1800–1999. University of
Maryland. Retrieved October 5, 2004 from www.bsos.umd.edu/
cidcm/inscr/polity
Martin, M. O., Mullis, I. V. S., Beaton, A. E., Gonzalez, E. J., Smith,
T., & Kelly, D. L. (1997). Science achievement in the primary
school years. Chestnut Hill: TIMSS Study Center.
Martin, M. O., Mullis, I. V. S., Gonzalez, E. J., Gregory, K. D., Smith,
T. A., Chrostowski, S. J., et al. (2000). TIMSS 1999 international
science report. Chestnut Hill: TIMSS Study Center.
Martin, M. O., Mullis, I. V. S., Gonzalez, E. J., & Chrostowski, S. J.
(2004). TIMSS 2003 international science report. Chestnut Hill:
TIMSS and PIRLS International Study Center.
McDaniel, M. A. (2006a). State preferences for the ACT versus SAT
complicates inferences about SAT-derived state IQ estimates: A
comment on Kanazawa. Intelligence, 34, 601−606.
McDaniel, M. A. (2006b). Estimating state IQ: Measurement
challenges and preliminary correlates. Intelligence, 34, 607−619.
Meisenberg, G. (in press). In God's image? The natural history of
intelligence and ethics. Brighton: Book Guild.
Meisenberg, G., Lawless, E., Lambert, E., & Newton, A. (2005). The
Flynn effect in the Caribbean: Generational change of cognitive
test performance in Dominica. Mankind Quarterly, 46, 29−69.
Meyer, J. W., Ramirez, F. O., & Soysal, Y. N. (1992). World expansion
of mass education, 1870–1980. Sociology of Education, 65,
128−149.
Mullis, I. V. S., Martin, M. O., Beaton, A. E., Gonzalez, E. J., Kelly,
D.L., & Smith, T. (1997). Mathematics achievement in the primary
school years. Chestnut Hill: TIMSS Study Center.
Mullis, I. V. S., Martin, M. O., Gonzalez, E. J., Gregory, K. D.,
Garden, R. A., O'Connor, K. M., et al. (2000). TIMSS 1999
international mathematics report. Chestnut Hill: TIMSS Study
Center.
141
Mullis, I., Martin, M. O., Gonzales, E. J., & Kennedy, A. M. (2003).
PIRLS 2001 international report. Chestnut Hill: IEA.
Mullis, I. V. S., Martin, M. O., Gonzalez, E. J., & Chrostowski, S. J.
(2004). TIMSS 2003 international mathematics report. Chestnut
Hill: TIMSS and PIRLS International Study Center.
Murray, Ch. (1998). Income inequality and IQ. Washington: The AEI
Press.
O'Driscoll, G. P., Holmes, K. R., & O'Grady, M. A. (2002). 2002
index of economic freedom. Washington: Heritage Foundation and
Wall Street Journal.
OECD (2000). Literacy in the information age. Final report of the
International Adult Literacy Survey. Paris: OECD.
OECD (2003). Literacy skills for the world of tomorrow. Further
results from PISA 2000. Paris: OECD.
OECD (2004a). Learning for tomorrow's world. First results from
PISA 2003. Paris: OECD.
OECD (2004b). Problem solving for tomorrow's world. Paris: OECD.
Oesterdiekhoff, G. W. (2000). Zivilisation und Strukturgenese.
Frankfurt: Suhrkamp.
Pearl, J. (2000). Causality. Models, reasoning, and inference. Cam-
bridge: Cambridge University Press.
Piaget, J. (1932). Le jugement moral chez l'enfant. Paris: PUF.
Piaget, J. (1947). La psychologie de l'intelligence. Paris: Libraire
Armand Colin.
Ram, R. (2007). IQ and economic growth: Further augmentation of
Mankiw-Romer-Weil model. Economics Letters, 94, 7−11.
Richards, M. (2002). IQ and the wealth of nations. Intelligence, 30,
574−576.
Rindermann, H. (2006). Was messen internationale Schulleistungs-
studien? Schulleistungen, Schülerfähigkeiten, kognitive Fähigkei-
ten, Wissen oder allgemeine Intelligenz? [What do international
student assessment studies measure? School performance, student
abilities, cognitive abilities, knowledge or general intelligence?].
Psychologische Rundschau, 57, 69−86.
Rindermann, H. (in press). The g-factor of international cognitive
ability comparisons: The homogeneity of results in PISA, TIMSS,
PIRLS and IQ-tests across nations. Target article for Special Issue
on European Personality Reviews (European Journal of
Personality).
Rindermann, H. (submitted for publication). Relevance of education
and intelligence for the political development of nations:
Democracy, rule of law and political liberty.
Rindermann, H., & Meisenberg, G. (submitted for publication).
Relevance of education and intelligence at the national level for
non-economic welfare: HIV and AIDS.
Rogosa, D. (1980). A critique of cross-lagged correlation. Psycholo-
gical Bulletin, 88, 245−258.
Rushton, J. Ph., & Jensen, A. R. (2005). Thirty years of research on
Black–White differences in cognitive ability. Psychology, Public
Policy, and Law, 11, 235−294.
Sala-i-Martin, X. (1997). I just ran two million regressions. American
Economic Review, 87, 178−183.
Sanderson, W. C. (2004). Interactions between education and HIV:
Demographic examples from Botswana. In W. Lutz, W. C.
Sanderson, & S. Scherbov (Eds.), The end of world population
growth in the 21st century (pp. 227−263). London: Earthscan.
Schermelleh-Engel, K., Moosbrugger, H., & Müller, H. (2003).
Evaluating the fit of structural equation models: Test of
significance and descriptive goodness-of-fit measures. Methods
of Psychological Research — online, 8, 23−74.
Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of
selection methods in personnel psychology: Practical and
142 H. Rindermann / Intelligence 36 (2008) 127–142
theoretical implications of 85 years of research findings. Psycho-
logical Bulletin, 124, 262−274.
Schofer, E., & Meyer, J. W. (2005). The worldwide expansion of
higher education in the twentieth century. American Sociological
Review, 70, 898−920.
Schooler, C., Mulatu, M. S., & Oates, G. (1999). The continuing
effects of substantively complex work on the intellectual
functioning of older workers. Psychology and Aging, 14,
483−506.
Shadish, W. R., Cook, Th. D., & Campbell, D. T. (2002). Experimental
and quasi-experimental designs for generalized causal inference.
Boston: Houghton Mifflin.
Skuy, M., Gewer, A., Osrin, Y., Khunou, D., Fridjhon, P., & Rushton,
J. Ph. (2002). Effects of mediated learning experience on Raven's
matrices scores of African and non-African university students in
South Africa. Intelligence, 30, 221−232.
Stelzl, I., Merz, F., Remer, H., & Ehlers, Th. (1995). The effect of
schooling on the development of fluid and cristallized intelligence:
A quasi-experimental study. Intelligence, 21, 279−296.
Templer, D. I., & Arikawa, H. (2006). The Jensen and the Hunt and
Sternberg comments: From penetrating to absurd. Intelligence, 34,
137−139.
Transparency International (2004). Global corruption report 2004.
London: Pluto Press.
UNAIDS/WHO (2003). AIDS epidemic update. Geneva: WHO.
UNDP and Arab Fund for Economic and Social Development (2003).
Arab Human Development Report 2003. Building a knowledge
society. New York: UNDP.
UNESCO (2004). 2005 EFA Global monitoring report. Paris:
UNESCO.
Vanhanen, T. (1997). Prospects of democracy. London: Routledge.
Vanhanen, T. (2005). Measures of democracy 1810–2004. Finnish
Social Science Data Archive (FSD). Retrieved January 17, 2006
from www.fsd.uta.fi/english/data/catalogue/FSD1289
Volken, Th. (2003). IQ and the wealth of nations. A critique of Richard
Lynn and Tatu Vanhanen's recent book. European Sociological
Review, 19, 411−412.
Weber, M. (1988/1920). Gesammelte Aufsätze zur Religionssoziologie
I (Protestantische Ethik, Konfuzianismus und Taoismus). Tübin-
gen: J.C.B. Mohr. (Original work published 1920).
Weede, E. (2006). Economic freedom and development: New
calculations and interpretations. Cato Journal, 26, 511−524.
Weede, E., & Kämpf, S. (2002). The impact of intelligence and
institutional improvements on economic growth. Kyklos, 55,
361−380.
Whalley, L. J., & Deary, I. J. (2001). Longitudinal cohort study of
childhood IQ and survival up to age 76. British Medical Journal
(BMJ), 322, 1−5.
Whaley, S. E., Sigman, M., Neumann, Ch., Bwibo, N., Guthrie, D.,
Weiss, R. E., et al. (2003). The impact of dietary intervention on
the cognitive development of Kenyan school children. Journal of
Nutrition, 133, 3965−3971.
Whetzel, D. L., & McDaniel, M. A. (2006). Prediction of national
wealth. Intelligence, 34, 449−458.
Winship, Ch., & Korenman, S. (1997). Does staying in school make
you smarter? The effect of education on IQ in The Bell Curve. In B.
Devlin, S. E. Fienberg, D. P. Resnick, & K. Roeder (Eds.), Intel-
ligence, genes and success (pp. 215−234). New York: Springer.